Sodium-Calcium Exchanger 1 Regulates Epithelial Cell Migration via Calcium-dependent Extracellular Signal-regulated Kinase Signaling*

Background: Sodium-calcium exchanger (NCX1) regulates calcium in renal epithelial cells. Results: Na,K-ATPase β-subunit regulates NCX1 membrane localization and reduced NCX1 expression or its functional inhibition increases cell migration. Conclusion: NCX1 plays a pivotal role in activation of calcium dependent migration via calmodulin/PI3K/ERK. Significance: Identifying regulators of epithelial cell motility is important in establishing novel therapeutic targets in fibrosis and cancer. Na+/Ca2+ exchanger-1 (NCX1) is a major calcium extrusion mechanism in renal epithelial cells enabling the efflux of one Ca2+ ion and the influx of three Na+ ions. The gradient for this exchange activity is provided by Na,K-ATPase, a hetero-oligomer consisting of a catalytic α-subunit and a regulatory β-subunit (Na,K-β) that also functions as a motility and tumor suppressor. We showed earlier that mice with heart-specific ablation (KO) of Na,K-β had a specific reduction in NCX1 protein and were ouabain-insensitive. Here, we demonstrate that Na,K-β associates with NCX1 and regulates its localization to the cell surface. Madin-Darby canine kidney cells with Na,K-β knockdown have reduced NCX1 protein and function accompanied by 2.1-fold increase in free intracellular calcium and a corresponding increase in the rate of cell migration. Increased intracellular calcium up-regulated ERK1/2 via calmodulin-dependent activation of PI3K. Both myosin light chain kinase and Rho-associated kinase acted as mediators of ERK1/2-dependent migration. Restoring NCX1 expression in β-KD cells reduced migration rate and ERK1/2 activation, suggesting that NCX1 functions downstream of Na,K-β in regulating cell migration. In parallel, inhibition of NCX1 by KB-R7943 in Madin-Darby canine kidney cells, LLC-PK1, and human primary renal epithelial cells (HREpiC) increased ERK1/2 activation and cell migration. This increased migration was associated with high myosin light chain phosphorylation by PI3K/ERK-dependent mechanism in HREpiC cells. These data confirm the role of NCX1 activity in regulating renal epithelial cell migration.

cells induced by stromal cell co-culture caused an up-regulation of NCX1 transcript level among other genes involved in cell adhesion (11). Furthermore, inhibition of NCX1 activity by KB-R7943 down-regulated cell adhesion molecule ICAM1 and suppressed cell adhesion (12). NCX1 works in close partnership with Na,K-ATPase, by utilizing the sodium gradient generated by Na,K-ATPase to drive calcium efflux. Na,K-ATPase has also been shown to function as a motility and tumor suppressor (13,14) and is involved in the maintenance of epithelial polarity (15) and cell adhesion (16,17). Moreover, we previously reported that knockdown of Na,K-ATPase ␤ 1 -subunit (Na,K-␤) in MDCK cells led to mesenchymal phenotype accompanied by increased cell proliferation via activation of phosphoinositide-3 kinase (PI3K)/Akt and extracellular-signal-regulated kinase (ERK1/2) pathways (18).
In this study we demonstrate that MDCK cells with Na,K-␤ knockdown (␤-KD) showed reduced NCX1 protein expression leading to an increase in intracellular calcium. Furthermore, we provide evidence that Na,K-␤ interacts with NCX1 and regulates NCX1 membrane localization. The activation of ERK1/2 and enhanced cell migration in ␤-KD cells was calcium-dependent and could be reversed when NCX1 was overexpressed in ␤-KD cells. Increased intracellular calcium activated calmodulin/PI3K/ERK signaling leading to myosin light chain kinase/ Rho-associated protein kinase-dependent migration. Furthermore, inhibition of NCX1 in MDCK, LLC-PK1, and human primary renal epithelial cells (HREpiC) also led to activation of ERK1/2 and enhanced cell migration. Thus, our data reveal a novel role of forward mode NCX1 in regulation of cell migration in renal epithelial cells.

EXPERIMENTAL PROCEDURES
Cell Lines and Maintenance-DMEM supplemented with 10% fetal bovine serum, 2 mM L-glutamine, 25 units/ml penicillin, and 25 g/ml streptomycin was used to grow MDCK and LLC-PK1 cells. Similarly, MDCK-Na,K-␤1-KD and rescue cells (␤-KD/R) as described in Barwe et al. (18) were also cultured in supplemented DMEM. ␤-KD cells were maintained in 10 g/ml puromycin, and ␤-KD/R cells were maintained in 10 g/ml puromycin and 500 g/ml neomycin. Full-length canine NCX1, a kind gift from Dr. Kenneth Philipson, UCLA (19), was transfected in ␤-KD cells using the calcium phosphate transfection method, and NCX1 expressing ␤-KD cells were selected with 10 g/ml puromycin and 100 g/ml hygromycin post transfection. ␤-KD cells were also transfected with pWZL-neo ⌬-p85 (Addgene Plasmid #10888) and selected with 10 g/ml puromycin and 500 g/ml neomycin post transfection. The cells that survived the selection media were confirmed to express the transfected constructs and were utilized for the experiments. HREpiC purchased from ScienCell™ (Carlsbad, CA) were maintained as per the supplier's recommendations and treated with inhibitors as indicated.
Immunoblotting and Co-immunoprecipitation-Cells were lysed in a buffer containing 20 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM ␤-glycerol phosphate, 1 mM sodium vanadate, 1 mM phenylmethylsulfonyl fluoride (PMSF), 5 mg/ml anti-papain, leupeptin, and pepstatin. For detection of NCX1 protein, cells were lysed in the lysis buffer supplemented with 2% sodium dodecyl sulfate (SDS). Based on protein estimation, 50 or 100 g of cell lysates were resolved by SDS-PAGE and transferred onto nitrocellulose membrane. The immunoblots were blocked in 5% nonfat dried milk in Tris-buffered saline with 0.1% Tween 20 (TBST). Primary antibodies were diluted either in 5% bovine serum albumin (BSA) or nonfat dried milk in TBST and incubated overnight at 4°C. Secondary antibodies were diluted in 5% nonfat dried milk in TBST. Immunoblots were developed with chemiluminescent lightning system ECL or ECL Prime (GE Healthcare) according to the manufacturer's recommendations. TINA 2.0 software (Open Source Image Analysis Environment) was utilized for immunoblot quantification and image analysis.
Cell lysate corresponding to 1 mg of protein was precleared with Protein A Mag-agarose beads (GE Healthcare) and incubated overnight with control IgG, Na,K-␤, or NCX1 antibodies pre-coupled to Mag beads for 4 h. The beads were washed and separated by SDS-PAGE, and the proteins bound to the beads were immunoblotted as described above. For resolution of Na,K-␤ protein (55 kDa) from heavy chain IgG band (50 kDa), immunoprecipitates were treated with peptide N-glycosidase F (New England Biolabs, Ipswitch, MA) as described previously (13). The deglycosylated Na,K-␤ (32 kDa) appears as a single band.
Quantitative Real-time PCR Analysis-RNA was extracted by TRIzol, and cDNA was generated by iScript™ cDNA Synthe-sis kit (Bio-Rad) as per the manufacturer's instructions. Quantitative PCR was performed using the SYBR Green PCR Master Mix (Applied Biosystems) in a 384-well plate on a 7900HT Fast Real-time PCR system (Applied Biosystems). The following primers were used: Na,K-␤ (forward, TTACCCTTACTACG-GCAAGCTCCT; reverse, TTCAGTGTCCATGGTGAGGT-TGGT); NCX1 (forward, TTAGCCGTTGTGGCTCTCTT;  reverse, TGTAGACCATGGCCACAAAA); GAPDH (forward, GCTGTCCAACCACATCTCCTC; reverse, TGGGGCCGAA-GATCCTGTT). RNA levels were calculated by relative quantification (RQ) normalized to the endogenous control GAPDH. Samples were assayed in triplicate. 35 S Metabolic Labeling-Cells (200,000) were grown on 60-mm dishes for 24 h in complete DMEM. Cells were incubated in methionine and cysteine-free DMEM containing 1% FBS 2 h before pulsing with 0.5 mCi/ml Tran 35 S-label (PerkinElmer Life Sciences) for 20 min. Cells were washed twice with PBS and lysed on ice in a lysis buffer. Lysates were incubated with protein A-agarose beads coated 1 g/ml NCX1 antibody. Beads were washed and resolved by 10% SDS-PAGE and detected by fluorography using Typhoon Trio-phosphorimager (GE Healthcare).
Immunofluorescence and Confocal Microscopy-Cells were cultured on glass coverslips and fixed with either ice-cold methanol (Ϫ20°C) or 4% paraformaldehyde in PBS at room temperature. The coverslips were incubated overnight at 4°C with primary antibodies diluted in 1% BSA in PBS with 100 M calcium chloride and 1 mM magnesium chloride followed by Alexa-488™-, Alexa-546™-, or Alexa-633™-conjugated secondary antibodies and TO-PRO-3-Iodide (Life Technologies) for nuclear staining. The coverslips were mounted on glass slides with ProLong gold antifade reagent (Life Technologies). The images were captured using Leica TCS SP5 Confocal Microscope (Leica Microsystems, Buffalo Grove, IL).
Wound-healing Migration Assay-Cells (200,000) were seeded in a 6-well tissue culture plate 24 h before a wound was made by scratching across the bottom of the well with a pipette tip. Wounded cell monolayers were washed three times with phosphate-buffered saline to remove the detached cells. Culture dishes were returned to the incubator for recovery of the wound in the presence or absence of the indicated inhibitors or reagents in serum-free media for 16 h. The scratches were photographed using an inverted microscope under the same configuration at the start and end of the experiment. The images were used to calculate the distance migrated by the cell sheet. The rate at which the wound was closed was calculated using the formula, speed ϭ distance/time.
Electrical Cell-substrate Impedance Sensing (ECIS) Migration Assay-ECIS Model 1600R, Applied BioPhysics (Troy, NY) was utilized. 1 ϫ 10 5 cells in 200 l of media were plated on ECIS electrode arrays (8W1E), each array containing 8 individual wells, with an active gold electrode in the base, and allowed to incubate overnight. Cell impedance and resistance levels were continuously measured. The resistance gained steadily and plateaued when cell confluence was attained. At this point, the electrodes were supplied an AC signal of 1 A, between a small measuring electrode (250-m diameter) and a large counter electrode, resulting in a sharp drop in resistance. As the cells re-grew and spread on the small electrode, the resistance increased proportionally to the distance migrated by the cells. The rate of migration was calculated as a ratio of the area covered (49,062.5 m 2 ) in a given period of time (number of hours taken for the resistance to reach the plateau).
Calcium Measurements by Flow Cytometry-Cells (5 ϫ 10 5 ) were plated in a 6-well dish for 48 h. On the day of the experiment cells were detached without the use of trypsin using 1ϫ Hanks' buffer saline solution ϩ 0.5 M EDTA at 37°C for 10 min, centrifuged at 3000 rpm for 3 min, and resuspended in complete medium. 1 ϫ 10 5 cells were resuspended in Tyrode's buffer without calcium containing 0.5 M Fluo4-AM in pluronic acid (both from Life Technologies) and incubated at 37°C for 20 min. Tyrode's buffer containing calcium was added to the cell suspension to make the final calcium concentration 0.5 mM before data acquisition using flow cytometer, BD Accuri C6 flow cytometer (BD Biosciences). The fluorescence intensity was continuously monitored for 8 -10 min. 100 M ouabain was added to the cells 2 min after start, and changes in intracellular calcium were calculated as described previously (20).
Calcium Imaging-Cells (10,000) were plated on glass-bottom dishes (MatTek Corp., Ashland, MA) and cultured until confluent. Cell cultures were then incubated with fluorescent calcium indicator, FURA-2, AM (Life Technologies) at a concentration of 10 M in a Pluronic acid/DMSO mixture in a 37°C incubator for 30 min protected from light. The MatTek plates were washed twice with HBSS containing 140 mM NaCl, 5 mM KCl, 2 mM CaCl 2 , 10 mM glucose, 10 mM 2-[4-(2-hydroxyethyl)piperazin-1-yl]ethanesulfonic acid, pH 7.4. Cells were allowed to recover after loading in Hanks' buffer saline solution for 10 min before imaging. A xenon lamp equipped with quartz collector lenses was used to excite the cells, and the cells were imaged with a Nikon inverted microscope (Intracellular Imaging, Cincinnati, OH). The fluorescence intensity was measured continuously for 5 min to obtain baseline calcium values. The ratio of the fluorescence intensity emitted at 510 nm after excitation at 340 nm over excitation at 380 nm was used to calculate -fold change in intracellular calcium concentrations as described previously (21).
Alternatively, cells were incubated with 10 M Fluo-4 AM in a Pluronic acid/DMSO mixture for 20 min at 37°C. The plates were washed twice with Hanks' buffer saline solution with or without 2 mM CaCl 2 . After recovery for 15 min, the fluorescence intensity corresponding to cytosolic calcium was continuously measured by an LSM 710 confocal microscope system (Zeiss, Thornwood, NY). The amplitude of the response was calculated by subtracting basal fluorescence obtained during the initial 2 min from the maximum intensity of fluorescence after adding the 10 M KB-R7943.
Statistical Analysis-Paired t test was used to evaluate the differences between average of two groups using data from at least three independent experiments, and a p of Ͻ 0.05 was considered statistically significant.

NCX1 Protein Level Is Reduced in ␤-KD
Cells-MDCK cells with stable knockdown of Na,K-␤ using the RNA interference technique (␤-KD) and with specific rescue of Na,K-␤ (␤-KD/R), MAY 15, 2015 • VOLUME 290 • NUMBER 20 generated by introducing silent mutations within the shRNA recognition site in the Na,K-␤ cDNA, have been described previously (18). Immunoblot analysis confirmed 80% reduction in Na,K-␤ protein level in ␤-KD cells. These cells also showed 42% reduction in NCX1 protein level compared with MDCK cells when normalized to ␤-actin used as a loading control (Fig. 1A). ␤-KD/R cells with renewed Na,K-␤ expression had restored NCX1 protein, highlighting the specificity of Na,K-␤ in the regulation of NCX1. The reduction in NCX1 protein in ␤-KD cells was consistent with the reduction of NCX1 in ␤-KO hearts as we reported previously (22). Changes in the protein levels of other Ca 2ϩ transport proteins such as PMCA and SERCA were not observed in ␤-KO hearts (22). Similarly, PMCA and SERCA protein levels in ␤-KD cells were comparable to parental MDCK cells (Fig. 1B), indicating that Na,K-␤ specifically regulates NCX1 protein.

Sodium-Calcium Exchanger1 Controls Epithelial Cell Migration
To determine whether Na,K-␤ alters NCX1 transcription, the mRNA level of NCX1 in ␤-KD cells was quantified. Compared with MDCK, ␤-KD cells did not show a significant change in NCX1 mRNA (Fig. 1C), suggesting that Na,K-␤-mediated NCX1 regulation is post-transcriptional. It has been reported previously that Na,K-␤ regulates the rate of protein synthesis of its binding partner Na,K-␣ (23). Therefore, we tested NCX1 protein synthesis rate in ␤-KD cells by performing 35 S metabolic labeling. The newly synthesized NCX1 protein detected by fluorography was comparable in MDCK, ␤-KD, and ␤-KD/R cells (Fig. 1D), indicating that Na,K-␤ does not alter the rate of synthesis of NCX1.
Na,K-␤ Interacts with NCX1 and Targets NCX1 to the Cell Surface-Next, we determined the membrane localization of NCX1 using cell surface biotinylation assay. Na,K-␤ on the cell surface was reduced by 85% in ␤-KD cells. The NCX1 level on the cell membrane was drastically diminished (by 89%) ( Fig.  2A), although the NCX1 total protein level was reduced only by 42% (Fig. 1A). NCX1 membrane expression was restored in ␤-KD/R cells, highlighting the specificity of Na,K-␤ in targeting NCX1 to the membrane. The cell surface level of occludin, a well known tight junction protein, remained the same and was used as a control for successful biotinylation and equal loading. Immunofluorescence analysis confirmed that NCX1 staining  was diminished in ␤-KD cells. Moreover, whereas NCX1 was localized to the membrane in MDCK and ␤-KD/R cells, such localization was not prominent in ␤-KD cells (Fig. 2B). Immunofluorescence analysis also showed that Na,K-␤ and NCX1 co-localize on the cell surface in MDCK and ␤-KD/R cells. To test the possibility that Na,K-␤ interacts with NCX1, a co-immunoprecipitation analysis was performed. Na,K-␤ was captured by anti-NCX1 immunoprecipitation, and NCX1 was detected in Na,K-␤ antibody immunoprecipitates (Fig. 2C), indicating that Na,K-␤ associates with NCX1.
␤-KD Cells Have Reduced NCX1 Activity and High Baseline Level of Intracellular Calcium-NCX1 activity is required for ouabain-induced increase in intracellular calcium (24). Therefore, the effect of ouabain on intracellular calcium was tested to determine whether NCX1 function is altered in ␤-KD cells. Treatment of MDCK and ␤-KD/R cells with 100 M ouabain elicited an immediate increase in intracellular calcium shown by the spike in fluorescence intensity above the threshold, i.e. the percentage of response to ouabain treatment above the baseline calcium level (Fig. 3A). However, the percentage of response to ouabain treatment was 50% lower in ␤-KD cells compared with MDCK cells, suggesting that ␤-KD cells have reduced NCX1 activity, consistent with reduced NCX1 protein.
NCX1 is the major Ca 2ϩ extrusion mechanism in kidney. To test whether reduced NCX1 affects calcium concentration in ␤-KD cells, the baseline intracellular free calcium levels were measured by Fura-2AM ratiometric imaging. The baseline intracellular calcium was 2.1-fold higher in ␤-KD cells compared with MDCK cells (Fig. 3B). The baseline calcium level in ␤-KD/R cells was similar to MDCK cells. Thus, the baseline intracellular calcium levels were inversely proportional to NCX1 protein levels.
␤-KD Cells Migrate Faster and Require Calcium for Increased Migration Rate-Because enhanced intracellular calcium can increase cell migration and Na,K-␤ is a motility suppressor (13,15), the rate of cell migration in ␤-KD cells was quantified by two different techniques. In a scratch wound healing assay, ␤-KD cells (13.6 Ϯ 1.5 m/h) showed a 1.8-fold higher migration rate than MDCK (8.2 Ϯ 0.5 m/h) and ␤-KD/R (7.8 Ϯ 0.7 m/h) cells (Fig. 4, A and B). A wound-healing assay was performed in the presence of calcium chelator, BAPTA-AM, to determine if the enhanced migration rate in ␤-KD cells was calcium-dependent. Treatment of ␤-KD cells with BAPTA-AM reduced the migration rate (5.6 Ϯ 0.3 m/h) (Fig. 4, A  and B), suggesting that migration in ␤-KD cells is calciumdependent.  In ECIS wound-healing assay, the resistance of cells grown on electrodes is continuously monitored. A wound is created by the application of high current to the cell-covered electrode resulting in cell death and causing a subsequent drop in the resistance. The rate of migration is calculated by determining how quickly the resistance reverts to the levels of the cell-covered electrode due to the migration of cells on to the electrode (25). ␤-KD cells showed a 2.4-fold higher migration rate when compared with parental MDCK cells (Fig. 4C). The differences in fold increase in the rate of migration could be due to the variances in the sensitivity of the equipment employed for migration analysis and the changes in local cell density as described previously (26). Thus, ␤-KD cells migrated faster in both assays, supporting the reliability of the ECIS wound-healing assay.
Cell Migration in ␤-KD Cells Requires Calcium-dependent ERK1/2 Activation-We previously demonstrated that ␤-KD cells have high ERK1/2 and Akt activation mediated by PI3K (18). These signaling pathways are known to play significant roles in inducing cell migration (27)(28)(29). Therefore, to determine which of these signaling molecules are regulated by intracellular calcium, MDCK, ␤-KD, and ␤-KD/R cells were treated with BAPTA-AM, and the phosphorylation status of ERK1/2 and Akt was determined by immunoblot analysis. BAPTA-AM treatment of ␤-KD cells reduced phosphorylated ERK1/2 to levels similar to MDCK cells (compare lanes 2 and 5), but the Akt phosphorylation remained high (Fig. 4D), indicating that ERK1/2 activation was calcium-dependent and was likely involved in the regulation of cell migration.
To determine the role of ERK1/2 and Akt activation in cell migration, ␤-KD cells were treated with specific inhibitors of these signaling pathways. Treatment with inhibitors of MEK1/2 (upstream regulator of ERK1/2), PD98059 and U0126, reduced the migration rate of ␤-KD cells from 9.5 Ϯ 0.9 m/h to 4.7 Ϯ 0.7 m/h and 5.7 Ϯ 1.1 m/h respectively, suggesting a significant role for ERK1/2 in migration (Fig. 5A).

ERK1/2 Activation and Cell Migration in ␤-KD Cells Is
Calmodulin-and PI3K-dependent but Akt-independent-As we showed previously that ERK1/2 phosphorylation in ␤-KD cells is PI3K-dependent (18), we tested the role of PI3K in ␤-KD cell migration. PI3K inhibition by LY294002 reduced phosphorylated ERK1/2 (47.9 Ϯ 1.3% of control) and also reduced the rate of migration (4.5 Ϯ 1.5 m/h) showing that PI3K is involved in inducing cell migration via its effect on ERK1/2. Because Akt is a downstream target of PI3K and is activated in ␤-KD cells, we tested whether Akt was also involved in the regulation of cell migration. Treatment with MK2206, a specific Akt inhibitor (10.9 Ϯ 1.2 m/h), did not alter the migration rate of ␤-KD cells (9.5 Ϯ 0.9 m/h) (p ϭ 0.355), indicating that migration in ␤-KD cells is Akt-independent. Immunoblot analysis of inhibitortreated cells confirmed the reduction in phosphorylated ERK1/2 by PD98059, U0126, and LY294002 (Fig. 5B). MK2206 did not change phosphorylated ERK1/2 levels significantly (83.9 Ϯ 1.9%, p ϭ 0.780). As expected, phosphorylation of Akt and p70S6 kinase, a downstream target of Akt, was reduced after LY294002 and MK2206 treatments (Fig. 5B).
To confirm the involvement of PI3K in cell migration, ␤-KD cells were transfected with a dominant negative mutant of PI3K p85 subunit lacking the binding site for the catalytic subunit p110 (␤-KD/⌬p85) (30). ␤-KD/⌬p85 cells had reduced Akt activation, consistent with suppression of PI3K signaling (Fig.  5D). Surprisingly, these cells migrated at rates similar to ␤-KD cells (Fig. 5C) and showed no reduction in ERK1/2 phosphorylation (Fig. 5D). This apparent discrepancy could be resolved by considering the fact that in addition to its well known lipid kinase activity, PI3K is also a protein kinase (31,32) that directly phosphorylates MEK resulting in ERK1/2 activation (33). The PI3K inhibitor LY294002 inhibits both activities of PI3K (34). ␤-KD/⌬p85 cells treated with LY294002 showed reduced migration and ERK1/2 phosphorylation (Fig. 5, C and D), indicating that the protein kinase activity of PI3K is involved in ERK1/2 phosphorylation and migration in ␤-KD cells.

Pharmacological Inhibition of NCX1 Induces PI3K-dependent ERK1/2 Activation and Promotes Cell Migration in Renal
Epithelial Cells-To determine if the functional inhibition of NCX1 is sufficient to activate ERK1/2 and induce migration, we blocked the Na ϩ /Ca 2ϩ exchange current with KB-R7943 that has been shown to inhibit both inward and outward calcium flux (38,39). Treatment of MDCK cells with 10 M KB-R7943 increased intracellular calcium in the presence or absence of calcium in the culture medium (Fig. 6, A and B). KB-R7943treated MDCK cells also showed a 2.8-fold increase in ERK1/2 phosphorylation, but the phosphorylated Akt level remained comparable with MDCK cells (Fig. 6C). Furthermore, KB-R7943 treatment in MDCK cells increased the migration rate by 1.5-fold (Fig. 6D), highlighting the role of NCX1 in the regulation of cell migration.
Overexpression of NCX1 in ␤-KD Cells Suppresses ERK1/2 Activation and Cell Migration-To confirm the role of NCX1 in the regulation of ERK1/2 and cell migration, we overexpressed NCX1 in ␤-KD cells (␤-KD/NCX1). Immunoblot analysis showed that the levels of NCX1 were 2.0 Ϯ 0.1-fold higher in ␤-KD/NCX1 cells (Fig. 8A). Higher NCX1 levels were accompanied by 75% reduction in ERK1/2 phosphorylation compared with ␤-KD cells. Unlike MDCK cells, ␤-KD cells do not show an increase in calcium response after KB-R7943 treatment. However, ␤-KD/NCX1 cells treated with KB-R7943 showed enhanced calcium response similar to MDCK, indicating that NCX1 activity was restored by overexpression of NCX1 (Fig.  8B). Finally, overexpression of NCX1 in ␤-KD cells suppressed migration rate in a wound-healing assay (Fig. 8C). Taken together, these data indicate that NCX1 expression in ␤-KD cells was sufficient to reduce calcium-induced ERK/12 activation and cell migration.

DISCUSSION
Our data show that reduced NCX1 function in ␤-KD cells was due to drastic reduction in NCX1 membrane localization.
In addition, the data suggest that Na,K-␤ associates with NCX1, and its abundance is affected in the knockdown of Na,K-␤ in ␤-KD cells. Furthermore, reduced NCX1 activity in these cells was associated with increased cell migration via calcium/calmodulin/PI3K-mediated ERK1/2 activation. Treatment of MDCK, LLC-PK1, and HREpiC cells with NCX1 inhibitor increased intracellular calcium, ERK1/2 phosphorylation, and enhanced migration, suggesting that the increased migration in ␤-KD cells is via suppression of NCX1 activity. This KB-R7943induced migration was associated with PI3K-and ERK1/2-dependent phosphorylation of MLC. Furthermore, ␤-KD cells with NCX1 overexpression restored NCX1 activity, resulting in reduced ERK1/2 phosphorylation and migration. A model summarizing the mechanism of migration induced by reduced NCX1 activity is outlined (Fig. 9).
Na,K-␤, the regulatory subunit of Na,K-ATPase, is important for co-translational transport and membrane insertion of Na,K-␣ (42). In addition, Na,K-␤ has been shown to interact with other membrane transporters such as Na ϩ -K ϩ -Cl Ϫ cotransporter (43) and large conductance Ca 2ϩ -activated K ϩ channels (44) and mediate membrane targeting of these transporters. In our study Na,K-␤ associated with NCX1, and Na,K-␤ knockdown reduced NCX1 protein, suggesting that it plays a similar role in aiding the trafficking and stabilization of NCX1 at the membrane. Because there is a reduction in the total NCX1 protein, whereas the rate of newly synthesized NCX1 is unaltered in ␤-KD cells, it is likely that NCX1 is targeted for degradation in ␤-KD cells.
NCX1 is a bidirectional ion counter transporter that can function either in the Ca 2ϩ efflux or Ca 2ϩ influx mode. NCX1  activity and the directionality of ion transport is driven by the membrane potential and the transmembrane electrochemical gradients of Na ϩ and Ca 2ϩ (45)(46)(47). The major function of NCX1 in the kidney is Ca 2ϩ extrusion, where it plays an important role in net transepithelial Ca 2ϩ absorption and is responsible for the majority of Ca 2ϩ extrusion (48). The importance of NCX1 in Ca 2ϩ efflux is evident by the rise in intracellular calcium in ␤-KD cells, which express reduced levels of NCX1. Although bi-directional NCX1 inhibitor KB-R7943 is more potent in inhibiting the reverse than the forward mode (49), its sensitivity is dictated by the splice variant of NCX1 expressed by the cells. The splice variant expressed in the kidney, NCX1.3 (50), is inhibited in its forward mode by KB-R7943 with IC 50 values of 2-3 M (51). Treatment with KB-R7943 in Ca 2ϩ -free medium also caused an increase in intracellular calcium (Fig.  6B). Taken together, the Ca 2ϩ efflux activity of NCX1 is involved in the regulation of cell migration in our study.
Many ion transporters regulate cell migration by maintaining housekeeping functions inside the cell such as membrane potential, cell volume, intracellular calcium concentration, and pH (52). NCX1 has been reported to promote the migration of a diverse group of non-epithelial cells such as endothelial cells, oligodendrocytes, myofibroblasts, microglia, and pancreatic cancer cells (53)(54)(55)(56)(57)(58)(59)(60). Inhibition or knockdown of NCX1 reduced intracellular calcium and prevented migration. This apparent contrast to our findings could be explained by the directionality of Ca 2ϩ flux by NCX1 in these cell types. This would suggest that NCX1 regulation of cell migration might be dependent on the directionality of ion transport. However, the underlying commonality in these reports and ours is that a rise in intracellular calcium was required for increase in cell migration. Migration is a calcium-dependent process as most of the regulators of motility that coordinate focal adhesion dynamics and interaction with cytoskeletal proteins are calcium-sensitive (61,62).
Ouabain, a specific inhibitor of Na,K-ATPase activity, leads to an increase in intracellular sodium. This increase reduces the driving force for the sodium influx through NCX1, subsequently resulting in an increase in intracellular calcium (24). NCX1 knock-out mice conclusively demonstrated that the ouabain-induced increase in intracellular calcium and enhanced cardiac contractility requires NCX1 activity (63). Therefore, ouabain sensitivity is dependent on NCX1 as well as Na,K-ATPase. Our previous studies showed that the Na,K-␤ KO cardiomyocytes and hearts with greatly reduced NCX1 were insensitive to ouabain-induced calcium transients and cardiac contractility (22). Consistent with these observations, ␤-KD with reduced NCX1 protein did not respond to ouabain, further confirming that NCX1 functions in the Ca 2ϩ efflux mode.
Migrating cells have a calcium gradient inside the cell with high calcium levels toward the retracting end and low calcium at the leading end (64), thus making it amenable to respond to calcium sparks that trigger forward movement. Other ion transporters involved in migration have a polarized distribution. The localization of NCX1 in a migrating cell has not been determined. Studies are currently under way in our laboratory to determine the role of NCX1 in directional movement.
PI3K is a dual specificity kinase that possesses serine/threonine protein kinase activity in addition to its well known lipid kinase activity (31,32). The lipid kinase activity is responsible for Akt phosphorylation, and the protein kinase activity mediates ERK1/2 activation by direct phosphorylation of MEK (33). In this study we observed calcium-dependent ERK1/2 activation in ␤-KD cells and after NCX1 inhibition; however, Akt was not dependent on calcium, indicating that the protein kinase activity of PI3K is involved in calcium-dependent ERK1/2 activation. A previous study has shown that direct binding of calmodulin to the SH2 domain of p85 subunit of PI3K results in increase of its lipid kinase activity (35); however, the effect on protein kinase activity was not tested. Increased lipid kinase activity in gain-of-function oncogenic mutants of PI3K p110 was accompanied by enhanced protein kinase activity (65). Therefore, it is tempting to speculate that calmodulin, by increasing the PI3K lipid kinase activity, also induces protein kinase activity of PI3K in renal epithelial cells. However, the exact mechanism by which calmodulin activates PI3K protein kinase activity remains to be determined.
Our previous studies have demonstrated that replenishing Na,K-␤ in highly motile Moloney sarcoma virus transformed-MDCK cells reduced cell motility (13,15) and that knockdown of Na,K-␤ prevented epithelial lumen formation via PI3K-dependent ERK1/2 activation (18). This is the first report showing that knockdown of Na,K-␤ in MDCK cells increased rate of cell migration and that increased intracellular calcium after NCX1 reduction was required for ERK1/2 activation and enhanced migration in ␤-KD cells. Thus, it is likely that reduced NCX1 is a downstream mediator of the ␤-KD phenotype.
The role of Na,K-␤ as a motility and tumor suppressor in conjunction with reduced expression of Na,K-␤ in several carcinomas has been established (13, 14, 66 -68). Similar to Na,K-␤ levels, NCX1 transcript level is also reduced in various subtypes of renal carcinoma (data not shown). Therefore, given the role of NCX1 in cell migration in epithelial cells, it is likely that reduced expression of NCX1 is also a prognostic indicator of carcinoma progression. Experiments are in progress to address the alternate functions of NCX1 in renal cancers.